Synthetic full and memory usage

[Pierpa]

Hi,
I have a veeam server 20GB memory 8 vcpu ( vmware virtual machine ) ( proxy maxtask 8 ) that, during the weekly synthetic full shows memory leak problems and does not complete the job. Therefore it cannot delete previous backups due to the fact that there is no a full backup apart the first Active full.
The backup server take care of about 100 vms, and I can see clearly a message on windows that states memory leak, and an error on veeam itself:

Code: Select all

2/5/2020 12:12:32 AM :: Agent: Failed to process method {Transform.CompileFIB}: memory no longer available. Err: 5001
Failed to read data from the file '[dd670_XXX.XXX.XXX.XXX] VEEAM_NEW:/VEEAM_BCK_XXX/XXX-YYY.vm-57D2020-01-10T220018_60FC.vib'. Offset: '0'.
Failed to open storage for read access. Storage: [dd670_XXX.XXX.XXX.XXX:VEEAM_NEW@/VEEAM_BCK_XXX/XXX-YYY.vm-57D2020-01-10T220018_60FC.vib].

The main question is: may I set on the veeam server vm a correct memory size in order to complete the synthetic full correctly and not resize the server with a useless too big memory ??
There is a formula I can apply to avoid this systematic error? Reducing the proxy maxtask from 8 to 6 can be an option ( I don't think so, since this is a memory issue ) or reducing the max storage repository task number can help ?


Thanks a lot for any suggestion
Pierpa

[foggy]

Hi Pierpa, please look at the system requirements outlined for every component here and keep in mind that having all-in-one installation (meaning that both proxy and repository components are running on the same server - do I understand right this is your case?) requires summing up the requirements for all the components. 20GB may not be enough for such a setup with 8 concurrent tasks.

[PetrM]

Hello,

One more idea is to check a block size at the level of backup job settings.
The usage of small block with large files may produce a large deduplication table and as result provoke high consumption of repository resources.

Thanks!

[Pierpa]

Hi,
reducing tasks seems to be a workaround to the memory leak. Thanks !!

Let's post another question. Suppose I have 2 proxies taking care of the backup job, the veeam backup server and another system.
During the synthetic full is the Backup server that crunches the incremental backups to merge them in a full or the second proxy works in collaboration with the backup server ??
In other words, during the synthetic full the second proxy stay idle or collaborates with the backup server?

Thanks a lot for any suggestion
Regards
Pierpa

[PetrM]

Hello,

The merge procedure is performed by Veeam Target Data Mover which is running on a repository (or gateway server).
Therefore, the backup server would be responsible for merge process only if backup jobs were pointed to a repository which is located on backup server itself.

All proxy servers stay in idle state during merge however if the same server has both proxy and repository (or gateway) roles: merge process will be done on this server.

Thanks!